Oak emerges from stealth to secure the identities of AI agents
Oak is applying continuous identity governance to employees, software and AI agents as enterprises create more non-human access paths.
AI agents are creating a new identity problem inside companies. They can access applications, retrieve data and take actions, but many enterprise security systems were built primarily around human employees.
What happened
Oak emerged from stealth and made its identity-management platform generally available. The company had raised $60 million in seed funding late last year from Accel, CRV, Greylock and other investors; the current event is the product launch, not a new financing round.
Oak maps how employees, software services and AI agents access enterprise applications. It then identifies unused, excessive or risky permissions and removes them continuously rather than relying only on periodic access reviews.
The product is aimed at a long-standing problem known as permission sprawl. Over time, users and systems accumulate access they no longer need, creating more opportunities for mistakes or abuse.
Why it matters
AI agents intensify that problem because companies may create large numbers of non-human identities that operate across several tools. An agent may need temporary access to email, cloud storage, databases or internal software, but traditional governance processes can struggle to track what it can do and whether those permissions remain necessary.
Oak is betting that identity security must become continuous and behaviour-aware. That is particularly important when an agent can act at machine speed and repeat an error across many systems.
The bigger picture
Agent security is developing into a distinct enterprise software category. Companies will need ways to authenticate agents, define their authority, monitor their activity and revoke access when risk changes.
Oak enters a crowded identity market, so its differentiation will depend on whether it can integrate deeply with existing systems and show that its automation reduces risk without disrupting legitimate work. The broader signal is clear: as enterprises adopt agents, identity governance becomes part of the core AI infrastructure stack.